The value of changing old customs: Bespoke software can be risky and expensive. But adaptable firms can benefit from off-the-shelf packages, says Nick Beard

Businesses are increasingly reliant on computers and computer software. This creates a great difficulty for information systems engineers today: complex business organisations demand custom-

designed software, yet this is risky and expensive to produce, and is prone to gradual failure.

Computer software is one of the most complex of all human artefacts. Its complexity brings problems - such as unpredictability and unmanageability. But complex though it is, software rarely exists in isolation. It is of little value unless it forms part of an information system: any organisation using computers has to have a set of policies, processes and procedures that determines how software is applied. So information systems are more complex than the software alone.

Any information system will contain an implicit model of the structure and workflow of the organisation that the software is supposed to serve. Such structures and workflows are rarely static, and the software has to change accordingly. As information systems age, they fail - organisations change faster than software, leaving the software out of date. To remain effective, software requires maintenance - yet it eventually becomes impossible to maintain. Each change introduces new problems as it fixes old ones. Too many changes, too many fixes: the result is a disaster of gradually failing code - software rust.

The solution is two-pronged: to use business process re-engineering and adaptable, off-the-shelf software.

There are traditionally two ways to obtain software: either pay someone to write bespoke software for you, or buy an off-the-shelf package. For common office systems, such as word-processors, a custom-made product is almost never justifiable. Many businesses, however, have requirements that warrant bespoke development. In addition, CASE-tool (Computer Assisted Software Engineering) salesmen and existing large computer departments within a company - both of which make a living by building custom software - are fighting rearguard actions against the trend away from bespoke.

Experiences of custom-built software are varied. Though they can be successful they are more commonly dismal failures. The disaster of the London Ambulance Service project was a very public example, but there are many, less sudden failures.

Even software that is initially successful can fail later. A major problem is the fate of software over time. Unless an organisation is unusually static, its software can become mismatched with the real needs of the business. Initially, small organisational changes can be accommodated through amendments to the software. However, changes can worsen the mismatch between implicit and real process, and the amount of 'patching' required to keep the software functioning increases. The result can be an 'implosion', where the system becomes unsupportable. This is a legacy of bespoke software designed without prior thought about systems optimisation or the potential for organisational change.

Unfortunately, many such systems were constructed when IT budgets were large enough to allow a company to indulge in ostentatious spending. (How many times were senior executives shown proudly around the humming air-

conditioned computer room?) The original software development is now frequently unaffordable, yet software development is not much cheaper than it was 10 years ago. The systems executive might appear to be in a cleft stick: has software become too expensive to maintain and too expensive to replace?

There is a way out: business process re-engineering (BPR), which is essentially starting from scratch, redesigning the ways businesses do things to render them more efficient, more effective and more customer-responsive. But BPR has implications beyond the implementation of information systems. It concerns reinventing existing organisations as a series of processes instead of separate tasks. Some of the notions involved are similar to those of other recently fashionable strands of management practice, such as Total Quality Management (TQM). For example, each process is re-evaluated from the customer's perspective. This provides a better way of obtaining a return on computer investment.

The best way to get value for money from computers is first to find ways to avoid using them. Leading advocate Michael Hammer, former Professor of Computer Science at MIT, now running consultants Hammer and Company, sums up the essence of BPR: 'Don't automate, obliterate]' Ask yourself: do you really need to do at all the things you are proposing to computerise?

The traditional response of organisations to complex processes is to demand custom - complex - software. An important benefit of BPR is that at the same time as making processes simpler, cheaper and more efficient, they can be redesigned to make use of tried-and-tested software. Users have always been concerned that packaged software never matches the quirks of their organisations. Instead of insisting on embedding these quirks in information systems, BPR argues for ridding the organisation of the quirks. Immediate one-off savings can be made - and longer-term efficiencies built into organisations.

Computer-based tools are now available which support this process. For example, BDF (Business Design Facility) from Texas Instruments enables organisations to develop detailed models of the way they operate, and to simulate the impact of process changes on the organisation's performance. Clive Mabey, services development director at TI, says: 'BDF makes it possible to understand the performance of the existing processes and assess alternatives for the future.'

BDF is a PC-based software. Using a series of easy-to-read diagrams, the BDF user can build models of his or her organisation, its activities and the relationships between them. The user can then analyse existing processes and design ways of improving business performance.

Re-engineering is a powerful but also a painful and potentially dramatic thing to inflict on a business. In Re-engineering the Corporation (Nicholas Brealey Publishing), Michael Hammer and James Champy describe the background, effects and methods of carrying out BPR. Properly conducted, it is the most likely method of obtaining real value from computing systems.

Dr Beard is director of information services at Health Care International, in Scotland.